Dynamic upscaling of decomposition kinetics for carbon cycling models

Abstract

The distribution of organic substrates and microorganisms in soils is spatially heterogeneous at the microscale. Most soil carbon cycling models do not account for this microscale heterogeneity, which may affect predictions of carbon (C) fluxes and stocks. In this study, we hypothesize that the mean respiration rate R at the soil core scale (i) is affected by the microscale spatial heterogeneity of substrate and microorganisms and (ii) depends upon the degree of this heterogeneity. To theoretically assess the effect of spatial heterogeneities on R, we contrast heterogeneous conditions with isolated patches of substrate and microorganisms versus spatially homogeneous conditions equivalent to those assumed in most soil C models. Moreover, we distinguish between biophysical heterogeneity, defined as the nonuniform spatial distribution of substrate and microorganisms, and full heterogeneity, defined as the nonuniform spatial distribution of substrate quality (or accessibility) in addition to biophysical heterogeneity. Four common formulations for decomposition kinetics (linear, multiplicative, Michaelis-Menten, and inverse Michaelis-Menten) are considered in a coupled substrate- microbial biomass model valid at the microscale. We start with a 2-D domain characterized by a heterogeneous substrate distribution and numerically simulate organic matter dynamics in each cell in the domain. To interpret the mean behavior of this spatially explicit system, we propose an analytical scale transition approach in which microscale heterogeneities affect R through the second-order spatial moments (spatial variances and covariances). The model assuming homogeneous conditions was not able to capture the mean behavior of the heterogeneous system because the second-order moments cause R to be higher or lower than in the homogeneous system, depending on the sign of these moments. This effect of spatial heterogeneities appears in the upscaled nonlinear decomposition formulations, whereas the upscaled linear decomposition model deviates from homogeneous conditions only when substrate quality is heterogeneous. Thus, this study highlights the inadequacy of applying at the macroscale the same decomposition formulations valid at the microscale and proposes a scale transition approach as a way forward to capture microscale dynamics in core-scale models.